Livelihoods, Basic Services, Social Protection and Perceptions of the State in Conflict-affected Situations Household Survey 2012

The sampling strategy was designed to select households that are relevant to the main research questions and as well as being of national relevance, while also being able to produce statistically significant conclusions at the study and village level. To meet these objectives, purposive and random sampling were combined at different stages of the sampling strategy. The first stages of the sampling process involved purposive sampling, with random sampling only utilized in the last stage of the process. Sampling locations were selected purposely (including districts and locations within districts), and then randomly households were selected within these locations. A rigorous sample is geared towards meeting the objectives of the research. The samples are not representative for the case study countries and cannot be used to represent the case study countries as a whole, nor for the districts. The samples are representative at the village level, with the exception of Uganda.

Sampling locations (sub-regions or districts, sub-districts and villages) were purposively selected, using criteria, such as levels of service provision or levels of conflict, in order to locate the specific groups of interest and to select geographical locations that are relevant to the broader SLRC research areas and of policy relevance at the national level. For instance, locations experienced high/ low levels of conflict and locations with high/ low provision of services were selected and locations that accounted for all possible combinations of selection criteria were included. Survey locations with different characteristics were chose, so that we could explore the relevance of conflict affectedness, access to services and variations in geography and livelihoods on our outcome variables. Depending on the administrative structure of the country, this process involved selecting a succession of sampling locations (at increasingly lower administrative units).

The survey did not attempt to achieve representativeness at the country /or district level, but it aimed for representativeness at the sub-district /or village level through random sampling (Households were randomly selected within villages so that the results are representative and statistically significant at the village level and so that a varied sample was captured. Households were randomly selected using a number of different tools, depending on data availability, such as random selection from vote registers (Nepal), construction of household listings (DRC) and a quasi-random household process that involved walking in a random direction for a random number of minutes (Uganda).

The samples are statistically significant at the survey level and village level (in all countries) and at the district level in Sri Lanka and sub-region level in Uganda. The sample size was calculated with the aim to achieve statistical significance at the study and village level, and to accommodate the available budget, logistical limitations, and to account for possible attrition between 2012-2015. In a number of countries estimated population data had to be used, as recent population data were not available.
The minimum overall sample size required to achieve significance at the study level, given population and average household size across districts, was calculated using a basic sample size calculator at a 95% confidence level and confidence interval of 5. The sample size at the village level was again calculated at the using a 95% confidence level and confidence interval of 5. Finally, the sample was increased by 20% to account for possible attrition between 2012 and 2015, so that the sample size in 2015 is likely to be still statistically significant. The overall sample required to achieve the sampling objectives in selected districts in each country ranged from 1,259 to 3,175 households.